PrObeD: Proactive Object Detection Wrapper

Authors: Vishal Asnani, Abhinav Kumar, Suya You, Xiaoming Liu

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments on MS-COCO, CAMO, COD10K, and NC4K datasets show improvement over different detectors after applying Pr Obe D.
Researcher Affiliation Collaboration Vishal Asnani Michigan State University asnanivi@msu.edu Abhinav Kumar Michigan State University kumarab6@msu.edu Suya You DEVCOM Army Research Laboratory suya.you.civ@army.mil Xiaoming Liu Michigan State University liuxm@cse.msu.edu
Pseudocode No The paper describes the stages and architecture of Pr Obe D but does not include any explicitly labeled 'Pseudocode' or 'Algorithm' block.
Open Source Code Yes Our models/codes are available at https: //github.com/vishal3477/Proactive-Object-Detection.
Open Datasets Yes Our experiments use the MS-COCO 2017 [44] dataset for GOD, while we use CAMO [39], COD10K [17], and NC4K [47] datasets for COD.
Dataset Splits Yes MS-COCO 2017 Val Split [44]: It includes 118,287 images for training and 5K for testing. COD10K Val Split [17]: It includes 4,046 camouflaged images for training and 2,026 for testing. CAMO Val Split [39]: It includes 1K camouflaged images for training and 250 for testing. NC4K Val [47]: It includes 4,121 NC4K images.
Hardware Specification Yes averaged across 1, 000 images, on a NVIDIA V 100 GPU.
Software Dependencies No The paper mentions using 'PyTorch [51]' but does not provide a specific version number for PyTorch or any other software component used in the experiments.
Experiment Setup No The paper describes the general training process (fine-tuning, end-to-end training, loss functions) but does not provide specific hyperparameters like learning rate, batch size, or number of epochs in the main text.